6. Perimeter Survey
The ice camp was located on a multi-year ice floe, with a region of flat first year
ice to the southwest where the runway was located. During the non-science
camp large, kilometre wide leads opened to the North and South of the ice camp.
These diverged and refroze. Later they became the sites of ridging and cracking,
working throughout the science camp. A ridge formed, from the refrozen lead ice
rafting onto thicker ice, to the north of camp on the evening of April 2nd. Part of
this ridge became the ‘ridge study site’. Ridges formed to the east and west of
the camp, just during the science camp. The ridge to the east of camp
continuously worked, forming cracks on it’s east side. The other two ridges where
less active. A lead (a.k.a. Pablo’s lead) close to the end of survey line 4, worked
continuously throughout the science camp, closing, ridging and reopening
several times.
These four active features (cracks, leads and ridges) surrounded the camp, at
times confined travel to within the area circumscribed by the leads and ridges.
Hence they were defined as the perimeter of our in-situ study area. In the second
week of camp, the first year ice the runway was situated on began to crack up,
forming cracks and ridges between the APLIS camp and “Pablo’s lead” at the
end of calibration transect line 3. There were numerous cracks in the first year
ice between camp and Pablo’s lead that were not surveyed.
In the short, 2 week, period of the ice camp it was not possible to do an in-depth
survey of all the deformation features within snowmobile distance from camp.
However, we were able to perform photographic surveys of all the active leads
and ridges that surrounded the camp. These surveys will be used to estimate
ridge volume and to pinpoint modes of dynamic activity we observe in satellite
and buoy data. To expand results to the 10km scale, we may make the
assumption that the ridges and leads at the “camp perimeter” were
representative of similar features within the 10km buoy array. Satellite imagery
will help us assess this assumption.
6.1 Photographic Survey of Dynamic ice
Andrew Roberts & Jacqueline Richter-Menge
A photographic survey of active sea ice features (leads, cracks & ridges)
accessible from the ice camp (within a 2 mile radius), was performed. On April
6th, 20 survey sites were set up. Each site was chosen as it contained an
interesting deformation feature that we expected would evolve dynamically. A
whippy flag was place to mark the photographic subject, and a second whippy
flag place 10 meters in front of the subject to mark the position of photographer.
This ensured that repeat photographs of the same site where comparable. A 2m,
10cm marked post was held at the subject point in each photograph.
An example of a set of photographs for one site is shown in fig 6.1. Sites were
visited every few days, if possible. Some site became inaccessible due to lead
activity between the sites and camp. In particular some sites to the south of camp
were only accessible once, as the first year ice in this vicinity started to break up
in the first week of camp. Some other sites were put in later than April 6th, to
compensate for the inaccessible sites. See appendix 6, table I for a list of
available photographs for each site. These photographs may be viewed at
http://research.iarc.uaf.edu/SEDNA/perimeter.php. Figure 6.2 shows the position
of sites in appendix 6, table I, relative to command and control at the ice camp.
Figure 6.1: Example of Photographic survey
for active site 2.0, which was located near
the end of calibration transect line 2.
Figure 6.2: Photographic Sites for dynamic ice survey. Active sites are labled clockwise from
Ridge Site. The Active Site at the Ridge Site is labled 1.0, and sites between survey lines 1 and 2
are labled 1.1, 1.2, 1.3. Sites at the end of each Calibration Transect are labled x.0 where x is the
calibration transect line number (1-6 clockwise from Ridge site), and a similar clockwise notation
is used for labelling active sites between calibration transects (x.1, x.2 etc.).
These photographs will be analysed by an undergraduate summer student, to
provide a detailed timeline of dynamic sea ice behaviour around the ice camp.
The photographs are useful as anecdotal evidence for the nature and magnitude
of specific deformation events we observed visually, in buoy track data and
RADARSat image analysis.
6.2 Ridge Block Size Surveys
Jennifer Hutchings & Alice Orlich
Photographic surveys of the ridges to the north and east of camp were performed
by Jennifer Hutchings, Pat McKeown, and Alice Orlich. We used a digital camera
body (Olympus Evolt-500) with a 30mm Seiko lens and OM-10 to O/S lens
adapter ring. In the centre of each photograph a 2m wooden post was placed
close to the ridge. The post was painted with 10cm intervals.
The intent of the photographs is to serve as a method of expediting block size
data collection. Presumably, the photos will archive more block size dimensions
than could be measured in-situ during a brief field campaign. The blocks can be
measured later by viewing the image on a computer, using the painted 2m pole
as a reference.
Back at UAF, Alice Orlich analyzed the photographs, measuring the dimensions
of each visible block or portion thereof, in the image. The dimensions recorded
were width(x), length(y), and depth(z) for each measurable block, as well as
block volume, when (x), (y), and (z) for a block made it possible. Ridge height
was also interpreted for the Ridge Site Calibration Survey. In relation to the ridge,
(x) is the axis of the ridge, (y) is the axis crossing the ridge, and (z) accounts for
the thickness of the block.
6.2.1 Ridge Site Calibration Survey
Alice Orlich
A series of photographs were taken on the 7th, 8th, and 10th of April at the first
flag away from the ridge on each of site transects R1 through R6 (Figure 5.4.To
estimate the accuracy of the photo-based data collection method, photographic
measurements were compared to actual in-situ block size measurements at the
Ridge Site (See 5.1 and Appendix 5, Table IV).
As each photo was analyzed, comments were recorded as to the quality of the
photo, perspective compared to others taken at same flag, and condition of ridge
material, i.e rubble, blocks, snow cover, and dynamics. Block dimensions were
taken only once for blocks appearing in multiple photographs so as not to include
duplicate data points. Duplicate photos were only used for reference, but were
counted into the Photo Efficiency Report (detailed below). The Ridge Site photo
collection includes surveys for each day, facing both North (away from APLIS)
and South (towards APLIS), so care was taken to not include the same blocks
from different directions, but it was discovered that the photos sometimes
revealed previously unseen blocks or provided better image quality due to varied
sun position. Few blocks were immediate to the 2m measuring pole, so most of
the measurements are estimated taking into account the distance from the pole.
The photos were arranged by Ridge Site Line to compare with the in-situ
measurements. It should be noted that the in-situ block measurements account
for only the blocks that fell along where the transect line crossed, whereas the
photos were taken at a distance away from the ridge, therefore including blocks
between each line. For example, a photo of Ridge Site Line 2 would include
blocks between Line 1 and Line 3. As noted earlier, duplicate blocks were
attributed to only one Line data set. Due to the increase in blocks available in the
photos, there was an average of 30% more data points found in the photos in
comparison to the in-situ measurements. We found the mean dimensions of all
blocks to vary between in-situ and photos by 42cm (x), 26cm (y), and 8cm (z). In
addition, a Photo Efficiency Report was created to determine the effectiveness of
the photo method to extract block size data. For evaluation purposes, a “data
point” refers to any block dimension or ridge height measurable in a photo. The
report concluded that of the 110 photos involved, 50.91% of them provided at
least 1 data point, while 20.91% yielded multiple data points, implying that
49.09% were either duplicates or of poor quality. The complete block size
comparison can be found in Appendix 6, Table Ia.
6.2.2 Perimeter Ridge Survey
Alice Orlich
This survey was conducted on two outlying ridges that appeared North (Jenny’s
Ridge) and East (East Shear Ridge) of APLIS. Due to the distance involved in
accessing the ridges, and the dynamic forces at work, the data collected was
only photos, with no in-situ measurements accompanying them. The sites were
visited on the 9th, 10th, 11th, 13th, and 14th of April. On any given day, one or both
ridges were photographed at various locations . Photographs were taken at
approximately 100m intervals along the ridges, the distance estimated by eye.
The photographer took the image standing 10m in front of the 2m pole.
The images were fewer in respect to distance covered when compared to the
Ridge Site Calibration Survey. The photo collections for each ridge illustrated a
great spectrum of features in both, considering that Jenny’s Ridge (JR) was a
compression ridge, where large cube-shaped blocks would be expected, and the
East Shear Ridge (ESR) was a shear ridge and therefore would be likely to
produce more rubble and indiscernible block forms. In fact, when investigating
the number of photos that included blocks, rubble, or snow cover or drifts,
surprisingly both ridges had nearly equal amounts of blocks(JR=81%, ESR=88%)
and rubble (JR= 36%, ESR=35%). Snow cover and drifts had a greater presence
in the East Shear Ridge, with 58% of the photos hampered by it, where JR had
only 12%. Very few blocks were immediate to the 2m measuring pole, so most of
the measurements are estimated taking into account the distance from the pole.
We found that the dynamic forces that created the ridges didn’t necessarily play
into the block size dimensions found in the measurements of this study. The
mean dimensions of all blocks vary between Jenny’s Ridge and the East Shear
Ridge by 15cm (x), 16cm (y), and 6cm (z). A Photo Efficiency Report was also
generated here, and it concluded that of the 50 photos involved, 32% of them
provided at least 1 data point, while 20% yielded multiple data points, implying
that 68% were either duplicates or of poor quality. The complete block size
comparison can be found in Appendix 6, Table Ib.
6.2.3 Future Work
After reviewing the data and statistics produced by the experiment, some new
methods are proposed. Although the in-situ measurements provide a great
sample for a ridge study, the method is time consuming and not entirely
complete, as it was discovered that some blocks can not be measured on all
axes due to overlap or submersion. It best serves as a quick, intense practice to
gather a sample of a portion of a ridge. The photo survey proved to be helpful in
yielding many additional data points, but which are subjective to the analyst’s
perception. We suggest that the two methods continued to be administered in the
field, but with a few modifications.
When performing any ridge study, like was done on the Ridge Site portion of
Jenny;s Ridge, in-situ measurements at transects lines can provide great
comparative reference data points. A future Perimeter Ridge Study could benefit
from this source of additional data. Also, it would be helpful if the blocks
measured in-situ on a transect were marked either by whippy flags or spray paint
for them to be easily identified for visual reference when analyzing the
accompanying photos. Of course, in-situ measurements should continue to be
collected as time warrants. Given the potential deformation that can be in
process while collecting data, it would be best if in-situ measurements along
ridge study areas can be planned to be a continuous effort. This would entail
scheduling ridge visits routinely to monitor creation or loss of blocks and features
for the entire ridge or designated portions, depending on length and camp staff.
The photo surveying technique was designed to capture more blocks per frame
than an in-situ transect line. Parallax error was considered by centering the pole
at the flagline of the Ridge Site Study transect lines. This, in addition to
positioning the pole at the blocks (rather than a few meters in front), provided for
the highest accuracy of block size measurements. To increase the accuracy of
blocks between flaglines, it is suggested that more photos be taken with the
intent to overlap edge features.
Lines
1
2
3
4
5
Ridge
Photos
X
X
X
X
X
X
X
X
X
Figure ??? Bird’s-eye view of how photos (X) along transect lines (1,2,3…) could create
overlap to ensure maximum block coverage with reduced parallax error.
As mentioned above, any blocks in the photos that have been previously
measured in-situ will aid in deducing photo block measurements. For this same
reason, it is suggested that for each photo taken, at least one block is measured
and marked by a member of the photo survey team as they progress along the
ridge. This insurance that every photo yields one data point will presumably
increase the number of data points interpreted during photo analysis.
The 2m pole was an efficient tool, both in regards to its transportability and
visibility in the photos. It could continue to be employed in future surveys, and
may be used in a more complicated mobile measuring system. If two like poles
were attached with lines strung between them at the 2m and 1m heights, and
markers of small surveying tape were set at 50cm spacings along the lines, a
mobile grid could be seen in each photo. The photo crew would include a
member setting the poles at what would be the edges of the photo while a
second member measures at least on block within the frame. The photographer,
as the third member, would be responsible for either pacing out the next photo or
determining where the features of the last photo could become the edge of the
next. In total, the three member crew is a size efficient enough to for safety,
accuracy and speed.
Figure ??? View from photographer’s position if two-pole reference measure grid is used.
Keeping in mind that meteorological conditions have a direct effect on the
outcome of the photo quality and data yield, it is recommended to take additional
photos with varying settings on the camera to ensure data recovery.
Figure 6.2: Location of ridges around the camp at the centre of the plot. Stars are the position of
block size photographs. The green line follows the compression ridge that formed to the north of
camp on April 2nd. The red line follows a shear ridge that formed, east of camp, on April 7th. The
blue line follows another shear ridge that formed, west of camp, on April 10th.
6.3 Stereo Photography Study
Participants: Cathleen Geiger, Scott Grauer-Gray, Robert Harris, Chandra
Kambhamettu, and Mani Thomas
6.3.1 Experiment Configuration
Using two free standing HP Photosmart 945 digital cameras (5.1 mega pixels)
held an arm’s length apart, a sequence of stereo images were taken of icescapes within a one mile radius of the APLIS ice camp (Appendix 6,Table II). A
cylinder of ice (Figure 6.3.1) that was a plug from one of the diver access hole
through the ice pack was measured and photographed in this manner as a
calibration source. At each station, a sequential shot of stereo pairs were taken
at sweeping angles about a particular feature. A very detailed set of shots was
taken of the calibration block (Figure 6.3.1), the main ridge studied in section 5
(Figure 6.3.2), the “Big Block”, the camp, the ice mass balance buoy, and the met
tower. This study was initiated as a bonus project to experiment with the ability to
simply collected stereo information of field observations as a means of providing
a complementary data resource for post-field analysis.
27cm
17cm
72 cm
29cm
33cm
41cm
156 cm
Figure 6.3.1: Calibration Block
1.3
Back
1.4
1.2
1.5
1.1
1.0
Front
2.3
2.4
2.2
2.5
2.1
2.0
3.3
3.4
3.2
4.3
4.4
3.5
3.1
4.2
3.0
4.5
5.3
6.3
5.4 5.5 6.4 6.5
4.1 5.2 5.1 6.2 6.1
4.0
5.0
6.0
Figure 6.3.2: Stereo shots at Station 13 – The Ridge Site
6.3.2 Example of ridge stereo photography analysis
We developed a stereo calibration and rectification system to produce 3D
reconstruction of some portions of stereo imaged data of ice. Three of the
example results are presented below. In figures 6.3.3 to 6.3.5, A and B represent
Stereo Left and Right images and C shows the reconstruction result. We are
currently designing advanced computer vision algorithms to work up on the
remaining portions of the stereo imagery having low texture, hence posing
severe challenges to stereo analysis process.
A
B
Figure 6.3.3: Stereo photography of a single block
C
A
B
C
Figure 6.3.4: Stereo photography of a ridge portion
A
B
Figure 6.3.5: Duplicate stereo photographs for ridge in figure 4.
C